A Comparative Evalution on the Prediction Performance of Regression Algorithms in Machine Learning for Die Design Cost Estimation

Abstract

Abstract: In the automotive industry, accurate estimation of mold costs is of great importance for businesses to maintain a competitive advantage and effectively manage costs. Traditional methods of predicting mold costs are time-consuming and prone to errors. Therefore, machine learning techniques, particularly regression algorithms, offer an innovative approach to mold cost estimation. This study aims to comparatively evaluate the performance of machine learning regression algorithms used in predicting mold costs in the automotive industry. Different types of regression algorithms, including Linear, Ridge, Lasso, K-Nearest Neighbors (KNN), Decision Tree, Random Forest, Support Vector Machine (SVM), Extreme Gradient Boosting, and Light Gradient Boosting Machines, were considered, and their performances on predicting mold costs and error rates were compared. The Random Forest Regression yielded the highest prediction accuracy at 98.197%.